I wouldn't characterize it as a "race to the bottom", although there's no doubt that there has been enormous price competition in this still-small niche market.
The cost of entry into the music software business is extremely low: basically you just need a computer and an internet connection. Well, and some rather arcane skills. But the skillset isn't out of reach for anyone already adept at some other category of realtime data processing, such as game programming.
Consequently, there is a growing cottage industry of one-man operations working out of their home offices, many in low-cost locales such as Eastern Europe, India and Turkey. Those guys can sell their products for 50 bucks a pop, and $49 of that will be pure profit.
Of course, there are many strata of skill levels among any group of software engineers. At the pinnacle of that hierarchy you've got a few lone wolves like Vojtech Meluzin and Aleksey Vaneev, true experts who choose not to be anyone's employee. But you'll find far more bit-flipping superstars anonymously drawing paychecks at traditionally-structured companies such as Waves, FabFilter, Sony and Avid. And make no mistake, a couple of them work at iZotope, too.
I found one example of iZotope's competence when I did a survey of harmonic exciters. There are a lot of them, and nearly every one of them suffers from the same problem: aliasing. But not Ozone. You simply cannot make any of Ozone's modules alias. And yet, if you look around Ozone's UI for the oversampling options, you won't find any. While any other dynamics processor requires you to manually guess at an appropriate oversample multiplier, and punish you for forgetting to set it before rendering, Ozone just quietly handles that automatically.
I have no reservations about lumping iZotope products with "the good stuff". Happily for us, it's good stuff we can now afford. Are they in financial trouble? It's possible, but I'm inclined to think it's just a matter of realistically adjusting price points in an evolving industry.